weathermodelsfandomcom-20200213-history
GFS
The Global Forecast System, mainly called GFS, is a global numerical weather prediction computer model run by NOAA. The model is run four times a day and is only the only one of the three main weather models that updates this. It also produces forecasts up to 16 days out however it is arguable that only up to the 5 or 7 day range is accurate. How it works The model runs from 0 hours to 192 hours in higher resolution. The resolution of the grid is internally about 40 km and for the continent map display 1 degree (about 100km). Then beyond 192 hours is run in a lower resolution. History The GFS model was originally called the Aviation Model (AVN) until its name was changed to GFS when the Aviation model was combined with the Medium Range Forecast Model (MRF). Known Problems It is widely known that the GFS can overdo pressure systems. One of the main model biases with the GFS is that the model tends to suffer from convective feedback. Convective feedback is when the model develops a local bulls eye of convection and heavy QPF over a very short distance. This convection could cause a large amount of latent heat to be released into the atmosphere, so much that the model has to compensate for the release, and produce a maximum of vertical velocity around 500mb over the QPF bulls eye. This also results in the development of a local upper-level jet maximum and a small intense bulls eye of vorticity. This can wreck havoc for a forecaster trying to use the model to predict where convection may develop. The issue with convective feedback happens primarily east of the Front Range and west of the Appalachian Mountains during the warm season. The GFS model also tends to have a dry bias north of areas where the model has predicted over 2” of QPF in a six hour period. This bias is usually found east of the Front Range and west of the Appalachians during the warm season. The bias also relates itself to the convective feedback issues because the QPF that is produced from convective feedback tends to block northward advection of moisture. The model also has a tendency to overdue the aerial coverage of very light precipitation (<.10”). This is common throughout the whole year across the whole domain of the model. In forecasting QPF, the model also has a tendency to vary significantly from model run to model run with QPF amounts. The model seems that it lacks continuity in QPF forecasts when compared to the ETA model. However, on the reverse side, the QPF verification on the GFS model has been proven to be better than the QPF verification on the ETA model, especially on forecasts that are 36 hours out in advance or greater. Reliability GFS is the only global weather model that is available for free in the public domain. Many weather websites across the world use it's data. The GFS also provides very good short range forcasts for many weather types something which many weather models do not offer. Performance In a experiment that took place over the course over nearly 3 months in 2011 the GFS peformed above average than most other models between 24 hours and 72 hours. However from 96 hours to 144 hours it performed mostly around average and at times above average than other weather models so overall it came second in the experiment. Sources http://en.wikipedia.org/wiki/Global_Forecast_System http://www.personal.psu.edu/adb241/eWallTutorial/models/gfs.htm Category:Global models